I have done a multiple regression analysis and I would like to determine the impact of each parameters on the model. It's the first time I do it on GraphPad and I don't find the way to do it.
This is a subject-matter question, not a statistical question.
You don't find importance in the data, no matter what tools you use. Importance relates to the meaning you give to the parts of a model, in relation to what aspect of the world you try to model and how you model it.
After obtaining the model you can check the t-test of each parameter. Any one with a p-value less than or equal to the significance level is deemed to have significantly affected the model. Otherwise it is deemed otherwise.
I would add some informations : I would to explain a performance by two different physiological characteristics that are correlated with my performance. The model has a r=0.83 with p
You can apply the partial F test, by comparing the F ratio associated with a model containing k-1 regressors and kth regressor as one of the two considered input variables, with another model containing the same k-1 regressors and the kth regressor as the other considered input variable.
The model with higher F ratio indicates that the particular kth regressor associated with this model is more significant given this precise set of remaining k-1 regressors.
Im not familiar with that software. I suggest you consider the meaning of “most important.” In a prediction context, i suppose you might look at increment to R-sq or increment to a model fit index, such as Bic or AIC. In a causal context, you have to consider the link between theory and model.
I agree with Jochen Wilhelm and Lawrence Raffalovich . I don't think GraphPad will help you. You might consider adaptive lasso which will not answer your question but will give you an optimal prediction equation given the the predictors that you started with(because of it's oracle property). You may wish to look at this link and it's references: